ISHA SUDHIR(STM20CS032)
NANDANA M(STM20CS041)
RAJATH P(STM20CS048)

Current methods compare the performance of eight pretrained models, but
they have limitations in generalization and detecting all forgery techniques.
We propose adopting suitable version of EfficientNet to improve accuracy
and versatility in forgery detection.
Aims to enhance authenticity assessment in the digital realm, especially on
previously unseen data where new forgery techniques may emerge.
Various versions of EfficientNet is known for its balance between model
size and performance.